添加ReadAllFromText转换时管道失败

Question

我正在尝试在Apache光束中运行一个非常简单的程序来测试它是如何工作的。

import apache_beam as beam


class Split(beam.DoFn):
    def process(self, element):
        return element


with beam.Pipeline() as p:
    rows = (p | beam.io.ReadAllFromText(
        "input.csv") | beam.ParDo(Split()))

在运行这段代码时，我得到以下错误

.... some more stack....
 File "/home/raheel/code/beam-practice/lib/python2.7/site-packages/apache_beam/transforms/util.py", line 565, in expand
    windowing_saved = pcoll.windowing
  File "/home/raheel/code/beam-practice/lib/python2.7/site-packages/apache_beam/pvalue.py", line 137, in windowing
    self.producer.inputs)
  File "/home/raheel/code/beam-practice/lib/python2.7/site-packages/apache_beam/transforms/ptransform.py", line 464, in get_windowing
    return inputs[0].windowing
  File "/home/raheel/code/beam-practice/lib/python2.7/site-packages/apache_beam/pvalue.py", line 137, in windowing
    self.producer.inputs)
  File "/home/raheel/code/beam-practice/lib/python2.7/site-packages/apache_beam/transforms/ptransform.py", line 464, in get_windowing
    return inputs[0].windowing
AttributeError: 'PBegin' object has no attribute 'windowing'

你知道这里出了什么问题吗？

谢谢

Answer 1

ReadAllFromText期望从文件的PCollection中读取，而不是将其作为参数传递。因此，在您的情况下，它应该是：

p | beam.Create(["input.csv"])
  | beam.io.ReadAllFromText()